Analysis · Health
When a Pregnancy Drug Risk Estimate Rests on Two Exposures
By AWEI · AI-compiled · Published · Analysis prepared · 1 source · doi.org
A review covering 58,461 patients can still produce a drug safety estimate supported by just two exposed pregnancies. That happened in a 2017 analysis of anti-epileptic drugs, illustrating why the size of a review tells only part of the story.
Areti Angeliki Veroniki and colleagues combined 96 studies examining congenital malformations and prenatal outcomes after exposure during pregnancy. Their search extended through December 2015. Most studies concerned epilepsy, and all but one were observational. The researchers used network meta-analysis to combine direct comparisons with indirect comparisons through shared treatments.
One striking result concerned topiramate and combined fetal losses. Its estimated odds ratio versus no drug exposure was 23.58, but the 95% credible interval stretched from 1.18 to 549.60. This describes enormous uncertainty around the estimate; it is not an absolute probability of pregnancy loss.
For that particular outcome, topiramate entered the network through one small study with only two exposed patients. The comparison with no exposure was entirely indirect. The study also had weaknesses in group comparability and follow-up. In another analysis that handled study designs differently, the interval included no difference. The finding was an uncertain association, not proof of causation or a reliable measure of its size.
The evidence had broader limitations. The authors reported that 81% of cohort studies did not control for important confounders, including factors such as epilepsy severity. Different doses were pooled because reporting was inadequate. Fetal loss reporting also varied, so the combined outcome did not consistently represent the same set of events across studies.
The practical lesson is to examine the evidence behind each drug and outcome separately. A large review can expose important safety signals while leaving particular comparisons unresolved. Here, interpreting the fetal loss estimate requires its uncertainty, underlying sample and study quality alongside the headline number.
Sources used for this article (1)
Direct links to the publisher reports used to prepare this article.
- Source 1
- Comparative safety of anti-epileptic drugs during pregnancy: a systematic review and network meta-analysis of congenital malformations and prenatal outcomes ['Areti Angeliki Veroniki', 'Elise Cogo', 'Patricia Rios', 'Sharon E. Straus', 'Yaron Finkelstein', 'Ryan Kealey', 'Emily Reynen', 'Charlene Soobiah', 'Kednapa Thavorn', 'Brian Hutton', 'Brenda R. Hem
